DISTRIBUTION ANALYSIS.

To analyse the distribution of a pottery 'ware' the procedure followed was to divide the sherds in each characterised ware into two functional classes; serving vessels (eg. Jugs, Bottles and Tripod Pitchers) and cooking vessels (eg. Cooking Pots, Bowls and Dripping Pans) and to try and assign the sherds to one of the following periods:-

• 10th Century
• 11th Century
• Early to mid-12th Century
• Late 12th to mid-13th Century
• Late 13th to mid-14th Century
• Late 14th to mid-15th Century
• Late 15th to mid-16th Century
• Late 16th to Early 17th Century.

If the ware had a short life-span (ie. within one of the periods defined above) then it was plotted onto one map showing all of the sites of that period with collections over 100 sherds. If the ware had a long life-span (as most late medieval wares did) then it was included on more than one map, making judgements on a case-by-case basis on the likely spread of occupation on the site and the evidence of nearby stratified sites of the same date. Collections producing more than 10% of the ware would be distinguished from those with less and a line drawn encompassing all these sites and excluding all others. This contour was then re-plotted onto a master map which was used for interpretation of the pattern of production and distribution.

The information represented by those collections with less than 10% of the ware was only analysed visually. If these sites formed a regular halo around the 10% frequency contour no further action was taken. If the sites formed a very diffuse halo this was noted for comment and similarly if the sites were irregularly distributed. A more objective analysis was not permitted because the the nature of the data.

The reliability of the contouring varies from period to period depending on the number of collections known and their location. From the 13th century onwards the contours overlap to such an extent that it is unlikely that further major wares remain to be found and the number and shape of the frequency contours is probably accurate. Before this date the data are much less reliable. Virtually every new area examined produces unpredictable results and probably only a fraction of the wares used in the study region is known. The distributions of the wares that are known are probably accurate in terms of scale but not in detail.